Project Tapestry Advances Sovereign AI Models via Global Consortium

The project’s recent technical breakthrough demonstrated a "consortium training" approach, where organizations across four geographically dispersed sites successfully trained and adapted AI models while keeping all underlying data local. This architecture addresses a primary hurdle in global AI development: the ability to build sophisticated systems without sacrificing data governance or national control. By modifying large language models to reflect the unique social nuances of Indian and Vietnamese contexts, the consortium has moved from conceptual design to functional, cross-border technology.

In New York, leadership met with Vietnamese officials, including General Secretary and President Tô Lâm, to integrate these capabilities into the nation’s AI strategy. Simultaneously, India deepened its involvement through the BharatGen initiative. Professor Ganesh Ramakrishnan, who leads the government-backed stack, recently coordinated a proof of concept between India and Australia. As the project shifts from its initial architecture to active deployment, stakeholders are preparing for a follow-up workshop in Mumbai on October 15, aiming to scale these collaborative frameworks across a growing network of international institutions.